Convert code base from boost::string_view to std::string_view

[carlhua]

Issue Description

The codebase currently does not make a lot of use of string_view. Mostly it shows up in the asio facing parts of the code. I tried converting those uses of boost::string_view to std::string_view a few months ago. It almost worked.

There’s a problem in boost 1.70 where in Windows the flag that tells beast to use std::string_view (not boost::string_view) does not work correctly. I understand that particular problem is fixed in Boost 1.71. So once we get to a place where the code base requires boost 1.71 or greater, then I think we can take on this task.

Once we have std::string_view in the code base there are a number of optimizations that can be performed. But I believe it should all start by moving to std::string_view first.

Exported from RIPD-1789

[intelliot]

@ckeshava could you think about this and let us know your recommendation?

Feel free to consult with @scottschurr as needed (he opened the original issue)

[ckeshava]

Hello,
As per my understanding, we are using Boost v1.77. So we should not run into the issue pointed out earlier. I will test for compatibility on Visual Studio 2019 after making these changes.

I believe std::string_view is almost always better than std::string const&. Do I need to think about other alternatives for boost::string_view? What do you think @scottschurr ?

[scottschurr]

Hi @ckeshava. Thanks for looking into this.

In terms of changes, my feeling is that, wherever we can, we should be replacing boost::string_view with std::string_view. The places that had problems before were boost::asio and boost::beast. Hopefully fixing those is now possible. I also had problems with soci wanting to use boost::string_view instead of std::string_view. It would be interesting to learn the state of the world on that front as well.

In terms of replacing std::string with std::string_view, that needs to be done very carefully with an eye toward lifetimes. So if we want to address that (and there are definitely places where that replacement is a win), that kind of change would want to be in one or more separate commits from the one described in the previous paragraph (just changing string_view types).

In case it's helpful, here is an outstanding review of what should (and should not) be done with string_view: https://www.youtube.com/watch?v=H9gAaNRoon4

I hope that helps. Let me know if you have further questions.

[thejohnfreeman]

We will need to finish eliminating string_views other than std::string_view if we want to upgrade to Boost 1.83 or later. The BEAST_USE_STD_STRING_VIEW macro that we have been relying on was removed in 1.83. We already removed boost::string_view in #4509, but 15 occurrences of boost::beast::string_view remain.